Gene sequencing data compression preprocessing, compression and decompression method, system, and computer-readable medium

Abstract

The present invention discloses a gene sequencing data compression preprocessing, compression and decompression method, a system, and a computer-readable medium. The preprocessing method implementation steps include: obtaining reference genome data; obtaining a mapping relationship between a short string K-mer and a prediction character c to obtain a prediction data model P 1 containing any short string K-mer in the positive strand and negative strand of a reference genome and the prediction character c in a corresponding adjacent bit. The compression and decompression methods relate to performing compression/decompression on the basis of the prediction data model P 1. The system is a computer system including a program for executing the previous method. The computer-readable medium includes a computer program for executing the previous method. The present invention can be oriented towards lossless gene sequencing data compression, provides fully effective information for a high-performance lossless compression and decompression algorithm for gene sequencing data.

What is claimed is: 
     
       1. A gene sequencing data compression preprocessing method, comprising the following implementation steps:
 1) obtaining reference genome data data ref ; 
 2) obtaining a mapping relation between any fixed length substring as the short string K-mer in the reference genome data data ref  and the prediction character c thereof, so as to obtain a prediction data model P 1  including any short string K-mer in the positive and negative strands of the reference genome and the prediction character c at the corresponding adjacent bit thereof. 
 
     
     
       2. The gene sequencing data compression preprocessing method as recited in  claim 1 , wherein the step 2) comprises the following implementation steps:
 2.1) sequentially extracting a fixed length substring in a positive strand S1 of the reference genome data data ref  as the short string K-mer to construct a positive strand short string set KS1 according to the designated space, wherein the positive strand S1 is the reference genome data data ref  of the original sequence; 
 2.2) sequentially extracting a fixed length substring in a negative strand S2 of the reference genome data data ref  as the short string K-mer to construct a negative strand short string set KS2 according to the designated space, wherein the negative strand S2 is a negative sequence complementary gene sequence of the reference genome data data ref ; 
 and between the negative sequence complementary gene sequence and the reference genome data data ref , the bases A and T are interchanged, and the bases C and G are interchanged; 
 2.3) generating a prediction data model P 1  corresponding to the reference genome data data ref  according to the positive strand short string set KS1 and negative strand short string set KS2, wherein the prediction data model P 1  contains the mapping relation between any short string K-mer in positive strand S1 and negative strand S2 and the prediction character c in the most possible adjacent bit obtained from statistics. 
 
     
     
       3. The gene sequencing data compression preprocessing method as recited in  claim 2 , wherein the step 2.3) comprises the following detailed steps:
 2.3.1) with respect to the positive strand short string set KS1, sequentially extracting the short string K-mer and constructing a positive strand prediction set KP1, wherein every element in the positive strand short string set KS1 has one corresponding tuple in the positive strand prediction set KP 1, and the tuple at least contains three types of information: short string K-mer, mark d from the positive strand, and the base letter c 0  in the adjacent bit of the positive strand S1; 
 2.3.2) with respect to the negative strand short string set KS2, sequentially extracting the short string K-mer and constructing a negative strand prediction set KP2, wherein every element in the negative strand short string set KS2 has one corresponding tuple in the negative strand prediction set KP2, and the tuple at least contains three types of information: short string K-mer, mark d from the negative strand, and the base letter c 0  of the element in the adjacent bit of the negative strand S2; 
 2.3.3) mapping the tuples in the positive strand prediction set KP1 and negative strand prediction set KP2 to the base letters A, C, G, T, counting any short string K-mer in the positive strand S1 and negative strand S2 and the base letters in the most possible adjacent bits obtained from statistics, obtaining the prediction data model P 1  containing any short string K-mer in positive strand and negative strand of the reference genome and the prediction data model P 1  of the prediction character c in the corresponding adjacent bit. 
 
     
     
       4. The gene sequencing data compression preprocessing method as recited in  claim 3 , wherein the step 2.3.3) comprises the following detailed steps:
 2.3.3.1) taking out every tuple (k-mer, d, c 0 ) one by one from the positive strand prediction set KP1 and the negative strand prediction set KP2, wherein k-mer is a short string K-mer corresponding to the tuple, d is the type of positive and negative strands, d=0 indicates the positive strand, d=1 indicates the positive strand, c 0  is the base letter of the adjacent bit corresponding to the short string K-mer of the corresponding tuple in the positive strand S1 or the negative strand S2; 
 2.3.3.2) using a preset mapping function to map a sub-tuple (k-mer, d) of every tuple (k-mer, d, c 0 ) taken out to a certain line of the integer set in a range of [0, L] so as to generate a 2D statistical table F containing L lines and 4 columns, and determining a corresponding line of a hit line of a corresponding base letter c 0  of the corresponding adjacent bit in the positive strand S1 or the negative strand S2 by virtue of the short string K-mer corresponding to the tuple in the tuple (k-mer, d, c 0 ), wherein L is a supremum of the integer set, and 0 is an infimum of the integer set; the number of base letters A, C, G, T corresponding to every value in the integer set and 4 columns of corresponding base letters A, C, G and T in the 2D statistical table F are counted; an element F i,c  in the 2D statistical table F stores the number of the base letters A, C, G, T corresponding to the sub-tuple (k-mer, d) with a value of i, in which a subscript i∈[0, L], c∈{A, C, G and T}; 
 2.3.3.3) traversing the 2D statistical table F from lines 0 to L, constructing the base letters corresponding to the element F i,c  , with the maximum value in every line to be a 1D character sequence as a prediction data model P 1  containing any short string K-mer in the positive strand and negative strand of the reference genome and the prediction character c of the corresponding adjacent bit, wherein the length of the prediction data model P 1  is L, L is the supremum of the integer set, the i(th) character P 1 [i] of the prediction data model P1 indicates the prediction character c of the short string K-mer corresponding to the tuple in the i(th) line of the integer set. 
 
     
     
       5. The gene sequencing data compression preprocessing method as recited in  claim 4 , wherein the step 2.3.3.2) of counting the number of A, C, G, T corresponding to every value in the integer set specifically refers to: when the sub-tuple (k-mer, d) of every tuple (k-mer, d, c 0 ) withdrawn is mapped to an integer set [0, L], with respect to four elements F i,c  in every line of the 2D statistical table, four count values F i,A , F i,C , F i,G , F i,T  are set, respectively;
 if the base letter c 0  of the short string K-mer corresponding to the tuple in the i(th) line of the integer set in corresponding adjacent bit of the positive strand S1 or negative strand S2 is hit as A, F i,A  in the i(th) line is plus 1; 
 if the base letter c 0  of the short string K-mer corresponding to the tuple in the i(th) line of the integer set in corresponding adjacent bit of the positive strand S1 or negative strand S2 is hit as C, F i,C  in the i(th) line is plus 1; 
 if the base letter c 0  of the short string K-mer corresponding to the tuple in the i(th) line of the integer set in corresponding adjacent bit of the positive strand S1 or negative strand S2 is hit as G, F i,G  in the i(th) line is plus 1; 
 if the base letter c 0  of the short string K-mer corresponding to the tuple in the i(th) line of the integer set in corresponding adjacent bit of the positive strand S1 or negative strand S2 is hit as T, F i,T  in the i(th) line is plus 1; 
 finally the numbers F i,A , F i,C , F i,T  of A, C, G, T corresponding to every value in the integer set are obtained. 
 
     
     
       6. The gene sequencing data compression preprocessing method as recited in  claim 2 , wherein the step 2.3) comprises the following detailed steps:
 S2.3.1) with respect to the positive strand short string set KS1, sequentially extracting the short string K-mer and constructing positive strand prediction set KP1, so that every element in the positive strand short string set KS1 has one corresponding tuple in the positive strand prediction set KP1, and the tuple at least contains three types of information: short string K-mer, mark d from the positive strand, and the base letter c 0  in the adjacent bit of the positive strand S1; 
 S2.3.2) with respect to the negative strand short string set KS2 sequence, extracting the short string K-mer and constructing negative strand prediction set KP2, so that every element in the negative strand short string set KS2 has one corresponding tuple in the negative strand prediction set KP2, and the tuple at least contains three types of information: short string K-mer, mark d from the negative strand, and the base letter c 0  in the adjacent bit of the negative strand S2; 
 S2.3.3) generating a training set by the short strings K-mer corresponding to the tuples in a positive strand prediction set KP1 and a negative strand prediction set KP2 and a base letter c 0  corresponding to the adjacent bit in the positive strand S1 or the negative strand S2 thereof, training a neural network model by the training set, and taking the trained neural network mode as the prediction data model P 1  containing any short string K-mer in positive strand and negative strand of the reference genome and the prediction data model P 1  of the prediction character c in the corresponding adjacent bit. 
 
     
     
       7. A gene sequencing data compression preprocessing system based on character prediction, comprising a computer system, wherein the computer system is programmed to perform the steps of the gene sequencing data compression preprocessing method as recited in  claim 1 . 
     
     
       8. A non-transitory computer-readable medium, on which a computer program is stored, the computer program allows the computer system to perform the steps of the gene sequencing data compression preprocessing method as recited in  claim 1 . 
     
     
       9. A gene sequencing data compression method, comprising the following implementation steps:
 1) traversing a gene sequencing data sample data to obtain a read sequence R with a length of Lr bit; 
 2) with respect to every read sequence R, selecting a original gene letter in k bit as an original gene character string CS 0 , generating k bit character string with fixed length based on the length of k bit as a sliding window sequence from the original gene character string CS 0 , as short string K-mer, determining a type d of positive and negative strands of the read sequence R based on the short string K-mer, and obtaining a prediction character c in the adjacent bit corresponding to every short string K-mer through the preset prediction data model P 1  to obtain a prediction character set PS with a length of Lr−k bit, wherein the prediction data model P 1  contains any short string K-mer in the positive and negative strands of the reference genome and a prediction character c in corresponding adjacent bit; 
 performing reversible computing by a reversible function after coding the Lr−k bit original gene letter and the prediction character set PS exclusive of k bit original gene letter in the read sequence R, wherein the output computing results coded by any pair of same characters are identical by virtue of the reversible function; 
 compressing and outputting the type d of the positive and negative strands of the read sequence R, k bit original gene letter and reversible computing result, as three data flows. 
 
     
     
       10. The gene sequencing data compression method as recited in  claim 9 , wherein the step 2) comprises the following implementation steps:
 2.1) traversing a read sequence R with a length of Lr from the gene sequencing data sample data, with respect to the read sequence R, selecting k bit original gene letter as an original gene character string CS 0 , generating fixed length substring based on the k bit length as the sliding window sequence from the original gene character string CS 0 , as the short string K-mer, and obtaining the read sequence short string set KR; 
 2.2) generating fixed length substring as the short string K-mer according to the sequence, determining the type d of positive and negative strands of the read sequence R based on the short string K-mer, and obtaining a prediction character c in corresponding adjacent bit of every short string K-mer to obtain a prediction character set PS with a length of Lr−k bit through a preset prediction data model P 1 , P 1 , wherein the prediction data model P 1  contains any short string K-mer in the positive and negative strands of the reference genome and a prediction character c thereof in the corresponding adjacent bit; 
 2.3) performing reversible computing by the reversible function after coding the Lr−k bit original gene letter and the prediction character set PS exclusive of k bit original gene letter in the read sequence R, wherein the output computing results coded by any pair of same characters are identical by virtue of the reversible function; 
 2.4) compressing and outputting the type d of the positive and negative strands of the read sequence R, the original gene character string CS 0  and reversible computing result, as three data flows; 
 2.5) judging whether the read sequence R in the gene sequencing data sample data is traversed, if not, jumping to step 2.1); otherwise ending and exiting. 
 
     
     
       11. The gene sequencing data compression method as recited in  claim 10 , wherein the step 2.2) comprises the detailed steps:
 2.2.1) sequentially extracting a short string K-mer with respect to the read sequence short string set KR, constructing a positive strand prediction sequence KP1 based on the short string K-mer, wherein any short string K-mer in a read sequence short string set KR has a corresponding tuple (k-mer, 0) in the positive strand prediction sequence KP1, k-mer is the short string K-mer, and 0 indicates supposing the short string K-mer from the positive strand; 
 2.2.2) obtaining the prediction character c corresponding to every tuple (k-mer, 0) in the positive strand prediction sequence KP1 through the prediction data model P 1 , P 1 , so as to obtain a positive strand prediction character sequence PS1 formed by all prediction characters c; the prediction data model P 1  contains any short string K-mer in the positive and negative strands of the reference genome and a prediction character c corresponding to the adjacent bit thereof; 
 2.2.3) sequentially extracting a short string K-mer with respect to the read sequence short string set KR, constructing a negative strand prediction sequence KP2 based on the short string K-mer, wherein any short string K-mer in the read sequence short string set KR has a corresponding tuple (k-mer, 1) in the negative strand prediction sequence KP2, k-mer is the short string K-mer, and 1 indicates supposing the short string K-mer from the negative strand; 
 2.2.4) obtaining the prediction character c corresponding to the adjacent bit through the prediction data model P 1  with respect to every tuple (k-mer, 1) in the negative strand prediction sequence KP2, and obtaining the negative strand prediction character sequence PS2 formed by all prediction characters c; 
 2.2.5) calculating an editing distance L1 between the positive strand prediction character sequence PS1 and Lr−k bit original gene letter exclusive of k bit original gene letter in the read sequence R, calculating an editing distance L2 between the negative strand prediction character sequence PS2 and the Lr−k bit original gene letter exclusive of k bit original gene letter in the read sequence R; 
 2.2.6) judging whether the editing distance L1 is less than L2, if yes, determining the type d of the positive and negative strands of the read sequence R to be a positive strand, and the positive strand prediction character sequence PS1 to be the prediction character set PS in Lr−k bit; otherwise, determining the type d of the positive and negative strands of the read sequence R to be negative strand, and the negative strand prediction character sequence PS2 to be the prediction character set PS in Lr−k bit. 
 
     
     
       12. The gene sequencing data compression method as recited in  claim 11 , wherein the prediction data model P 1  is a gene character string with a length of L, and the step 2.2.2) of obtaining the prediction character c in the adjacent bit thereof corresponding to every tuple (k-mer, 0) in the positive strand prediction sequence KP1 through the prediction data model P 1  comprises the following detailed steps:
 2.2.2.1) mapping every tuple (k-mer, 0) in the positive strand prediction sequence KP1 to a certain line of an integer set [0, L] using a mapping function corresponding to the prediction data model P 1 , P 1 , wherein L is a supremum of the integer set which is as long as the prediction data model P 1 , P 1 , and 0 is an infimum of the integer set; 
 generating a one-dimensional (1D) table T1 with a length of (Lr−k+1) according to a mapping result, wherein the i(th) element T1[i] in the 1D table T1 is respectively and sequentially stored and mapped to a value of the mapping function corresponding to the tuple (k-mer, 0) in the i(th) line of the integer set, i∈[0, Lr−k]; 
 2.2.2.2) obtaining a prediction character c corresponding to the adjacent bit from the prediction data model P 1  according to the value of the mapping function corresponding to each tuple of 1D table T1, and generating 1D character sequence PS1, so that the i(th) PS1 [i] value of the 1D character sequence PS1 is equal to the i(th) character P 1 [T1[i]] in the prediction data model P 1 , P 1 , and the i(th) character P 1 [T1[i]] in the prediction data model P 1  is the corresponding character c with the mapping function value of PS1 [i] corresponding to the tuple (k-mer, 0), wherein i∈[0, Lr−k], Lr is a read length of the read sequence R, and k is the length of the short string K-mer. 
 
     
     
       13. The gene sequencing data compression method as recited in  claim 12 , wherein the step 2.2.4) of obtaining the prediction character c in the adjacent bit thereof corresponding to every tuple (k-mer, 1) in the negative strand prediction sequence KP2 through the prediction data model P 1  comprises the following detailed steps:
 2.2.4.1) mapping every tuple (k-mer, 1) in the negative strand prediction sequence KP2 to a certain line of an integer set [0, L] using the mapping function corresponding to the prediction data model P 1 , P 1 , wherein L is a supremum of the integer set which is as long as the prediction data model P 1 , P 1 , and 0 is an infirnum of the integer set; 
 generating a 1D table T2 with a length of (Lr−k+1) according to a mapping result, wherein the i(th) element T2[i] in the 1D table T2 is respectively and sequentially stored and mapped to a value of the mapping function corresponding to the tuple (k-mer, 1) in the i(th) line of the integer set, i∈[0, Lr−k]; 
 2.2.4.2) obtaining the prediction character c corresponding to the adjacent bit thereof from the prediction data model P 1  thereof according to the value of the mapping function corresponding to each tuple (k-mer, 1) in the 1D table T2, and generating the 1D character sequence PS2, so that the i(th) PS2[i] value of the 1D character sequence PS2 is equal to an i(th) character P 1 [T2[i]] in the prediction data model P 1 , P 1 , wherein the i(th) character P 1  [T2[i]] in the prediction data model P 1  is the corresponding prediction character c with a mapping function value of PS2[i] corresponding to the tuple (k-mer, 0), i∈[0, Lr−k], Lr is a read length of the read sequence R, and k is the length of the short string K-mer. 
 
     
     
       14. The gene sequencing data compression method as recited in  claim 11 , which characterized in that, wherein the prediction data model P 1  is a neural network model which is trained in advance based on the short string K-mer in the reference genome and the corresponding base letter c 0  in the adjacent bit thereof;
 the step 2.2.2) of obtaining the prediction character c in the adjacent bit thereof corresponding to every tuple (k-mer, 0) in the positive strand prediction sequence KP1 through the mapping function of the prediction data model P 1  specifically refers to inputting every tuple (k-mer, 0) in the positive strand prediction sequence KP1 into the neural network model to obtain the prediction character c corresponding to the tuple (k-mer, 0); 
 the step 2.2.4) of obtaining the prediction character c in the adjacent bit thereof corresponding to every tuple (k-mer, 1) in the negative strand prediction sequence KP2 through the prediction data model P 1  specifically refers to inputting every tuple (k-mer, 1) in the negative strand prediction sequence KP2 into the neural network model to obtain the prediction character c corresponding to the tuple (k-mer, 1). 
 
     
     
       15. The gene sequencing data compression method as recited in  claim 9 , wherein an XOR function or a bit subtraction function is specifically applied for the reversible function in step 2). 
     
     
       16. The gene sequencing data compression method as recited in  claim 9 , wherein the compression in step 2) specifically refers to a compression using a statistical model and entropy coding. 
     
     
       17. A gene sequencing data compression system, comprising a computer system, wherein the computer system is programmed to perform the steps of the gene sequencing data compression method as recited in  claim 9 . 
     
     
       18. A non-transitory computer-readable medium on which a computer program is stored, the computer program enables a computer to perform the steps of the gene sequencing data compression method as recited in  claim 9 . 
     
     
       19. A gene sequencing data decompression method, comprising the following implementation steps:
 1) traversing gene sequencing data data c  to be decompressed to obtain a read sequence R c  to be decompressed; 
 2) with respect to every read sequence Rc to be decompressed, first decompressing and reconstructing the read sequence R c  to be decompressed as a positive and negative strand type d, an original gene sequence CS1 in the k bit and a reversible computing result CS2 with the length of Lr−k bit; 
 taking the original gene sequence CS1 in the k bit as an initial short string K-mer, and obtaining the corresponding prediction character c in the adjacent bit of the short string K-mer through a preset prediction data model P 1 , P 1 , wherein the prediction data model P 1  contains any short string K-mer in the positive and negative strands of the reference genome and a corresponding prediction character c in the adjacent bit thereof; 
 when one prediction character c is obtained, forming a new short string K-mer constituted by new prediction character c and the last k-1 bit of the short string K-mer, obtaining the new prediction character c by the preset prediction data model P 1 , and eventually obtaining a prediction character set PS with a length of Lr−k bit constituted by all prediction characters c; 
 performing the reverse computing for the coded reversible computing result CS2 and the prediction character set PS to obtain the decrypted result of the reversible computing result CS2 in the Lr−k bit by virtue of the inverse function of the reversible function; 
 combining the decrypted results of the original gene sequence CS1 in the k bit and the reversible computing result CS2 to obtain the original read sequence R corresponding to the read sequence Rc to be decompressed, and outputting the original read sequence R. 
 
     
     
       20. The gene sequencing data decompression method as recited in  claim 19 , wherein the step 2) comprises the following detailed steps:
 2.1) traversing gene sequencing data data c  to be decompressed to obtain a read sequence R c  to be decompressed; 
 2.2) decompressing and reconstructing the read sequence R c  to be decompressed to be a positive and negative strand type d, an original gene sequence CS1 in the k bit and a reversible computing result CS2 with a length of Lr−k bit; 
 2.3) taking the original gene sequence CS1 in the k bit as an initial short string K-mer, and obtaining a corresponding prediction character c in the adjacent bit of the short string K-mer through a preset prediction data model P 1 , P 1 , wherein the prediction data model P 1  contains any short string K-mer in the positive and negative strands of the reference genome and the corresponding prediction character c in the adjacent bit thereof; 
 when one prediction character c is obtained, forming a new short string K-mer constituted by new prediction character c and the last k-1 bit of the short string K-mer, obtaining the new prediction character c by the preset prediction data model P 1 , P 1 , and eventually obtaining a prediction character set PS with a length of Lr−k bit constituted by all prediction characters c; 
 2.4) performing the reverse computing for the coded reversible computing result CS2 and the prediction character set PS to obtain the decrypted result of the reversible computing result CS2 in the Lr−k bit by virtue of the inverse function of the reversible function; 
 2.5) combining the decrypted results of the original gene sequence CS1 in the k bit and the reversible computing result CS2 to obtain the original read sequence R corresponding to the read sequence R c  to be decompressed, and outputting the original read sequence R; 
 2.6) judging whether the read sequence R c  to be decompressed in the gene sequencing data sample data c  to be decompressed is traversed, if not, jumping to step 2.1); otherwise ending and exiting. 
 
     
     
       21. The gene sequencing data decompression method as recited in  claim 20 , wherein the step 2.3) comprises the following detailed steps:
 2.3.1) creating a window variable CS and a prediction character set PS corresponding to the short string K-mer, setting the initial value of the window variable CS as an original gene sequence CS1 in the k bit, creating an iteration variable j and setting the initial value as 0; 
 2.3.2) constructing the window variable CS, the type d of positive and negative strands of the read sequence Rc to be decompressed into a tuple (CS, d), mapping the tuple (CS, d) into an integer set [0, 1] by the mapping function, wherein L is the supremum of the integer set and equal to the length of a prediction data model P 1 ; 0 is the infimum of the integer set, and the prediction data model P 1  contains any short string K-mer in the positive and negative strands of the reference genome and a prediction character c thereof corresponding to the adjacent bit; 
 2.3.3) querying the i(th) P 1 [i] in the prediction data model P 1  with the function value obtained from the mapping function, as the prediction character c corresponding to the adjacent bit of the window variable CS, wherein i∈[0, L]; assigning the prediction character c to the j(th) bit in the prediction character set PS, wherein j∈[0, Lr−k], and Lr−k is the length of reversible computing result CS2; 
 2.3.4) after combining the last k-1 bit of the window variable CS and the prediction character c obtained currently, assigning to the window variable CS, and adding 1 to the iteration variable j; 
 2.3.5) judging whether the iteration variable j is greater than the length (Lr−k) of the reversible computing result CS2, if yes, jumping to the next step, otherwise, jumping to step 2.3.2); 
 2.3.6) outputting the prediction character set PS with the length of (Lr−k). 
 
     
     
       22. The gene sequencing data decompression method as recited in  claim 21 , wherein the step 2.3) comprises the following detailed steps:
 S2.3.1) creating a window variable CS and a prediction character set PS corresponding to the short string K-mer, setting the initial value of the window variable CS as the original gene sequence CS1 in the k bit, creating an iteration variable j and setting the initial value as 0; 
 S2.3.2) inputting the window variable CS to the prediction data model P 1  to obtain the prediction character c of the short string K-mer corresponding to the adjacent bit in the positive strand and negative strand of the reference genome, wherein the prediction data model P 1  is the neural network model which is trained in advance based on the short string K-mer in the reference genome and the base element c 0  corresponding to the adjacent bit thereof; 
 S2.3.3) assigning the prediction character c to the j(th) bit in the prediction character set PS, wherein j∈[0, Lr−k], and Lr−k is the length of reversible computing result CS2. 
 S2.3.4) after combining the last k-1 bit of the window variable CS and the prediction character c obtained currently, assigning to the window variable CS, and adding 1 to the iteration variable j; 
 S2.3.5) judging whether the iteration variable j is greater than the length (Lr−k) of the reversible computing result CS2, if yes, jumping to the next step, otherwise, jumping to step 2.3.2); 
 S2.3.6) outputting the prediction character set PS with the length of (Lr−k). 
 
     
     
       23. The gene sequencing data decompression method as recited in  claim 19 , which characterized in that, wherein the reversible function specifically refers to an XOR function or a bit subtraction function; An inverse function of the XOR function is the XOR operation, and an inverse function of the bit subtraction function is a bit addition function. 
     
     
       24. The gene sequencing data decompression method as recited in  claim 19 , wherein the decompression and reconstruction in step 2) specifically refer to a decompression and reconstruction using inverse algorithms of a statistical model and entropy coding. 
     
     
       25. A gene sequencing data decompression system based on character prediction, comprising a computer system, wherein the computer system is programmed to perform the steps of the gene sequencing data decompression method as recited in  claim 19 . 
     
     
       26. A non-transitory computer-readable medium, on which a computer program is stored, the computer program allows the computer to perform the steps of the gene sequencing data decompression method as recited in  claim 19 .